One more for the road · 8:34pm Jul 30th, 2014
Setting a daily record here.
After the last post, I got a bug in my brain about view/vote ratios (maybe I should call them VVR), and I decided to go through my account and find out some stats. First, a few things.
I figure it's only worth doing this for stories with over a thousand views. I have 22 23 of those. Second, the way it's calculated is adding upvotes and downvotes; using the total votes, you get a measurement of engagement, i.e., how many people had enough of a reaction to your story to actually cast a vote one way or another. So let's see what I got. (I rounded the decimals off, but these are totally in order.)
1) Princess Cadence Takes One for the Team- 1:6
2) Faun- 1:7
3) Snowdrop and Nyx Get Drunk and Make Out- 1:7
4) Epic Unicorn History- 1:9
5) My Full-Sized Goddess Horse- 1:10
6) Fall Down- 1:10
7) Dinky's First Kill- 1:11
8) Civil Twilight- 1:11
9) Treehouse- 1:11
10) Chrysalicksy- 1:11
11) A Lovely Apology- 1:12
12) Discord's Deli of Chaos- 1:13
13) Dance 'Til We're High- 1:13
14) Whisk- 1:16
15) Pinkie Pie, and Nothing More- 1:17
16) The Elements of Awesomery- 1:19
17) ACT OF WILL- 1:20
18) Fluttershy Goes to Fucking Narnia- 1:20
19) Things Rainbow Dash Doesn't Like- 1:21
20) Fluttershy Goes to Hell- 1:24
21) Twilight Sparkle Loses Her Wings- 1:24
22) Elements of Awesomery Side Stories- 1:25
23) Future Progressive- 1:40
First thing I note is this metric favors stories closer to a thousand views, and it favors newer stories. (Anyone remember when Fimfic used a five-star scale? I bet the switch from that to up/down has something to do with this.) Second, take a look at Epic Unicorn History. That's a long, ongoing story that's been around for a while, but it's fourth. I feel that this somehow paints it as a truly engaging story, beyond the three rated above it. I also feel like if you're under 1:10, you've made an engaging story (I've come to the conclusion lately that if you have x number of subscribers on a service, you can expect x/10 of them to do any particular thing, like say, buy an album.) I also also have come to the conclusion that under 1:6 is "truly remarkable". I'm not sure why I think that, it's just kind of been bandied around lately.
Anyway, this proves precisely jack all. There'd need to be a way to fit time spent on the site into it to really come up with a useful number. (And I keep thinking word count should be a factor, but I dunno.) What's the point of this? I LIKE MATH LET'S FIGHT
So, what, you add the votes to get your total, then divide it by the viewcount?
As time goes on, I think your ratio falls naturally as people who are only sort of mildly interested check into your story and are too meh about it to up or down vote it. You get the best ratio at the start when people are most enthusiastic.
Could have sworn I'd Thumbed and Faved Epic History. But I hadn't! FIXED!
My feed right now.
I use a 1:10 ratio of likes to views to figure out what stories to read. I completely ignore downvotes unless it's a significant amount, which I pretty much only find in clopfics. It's a pretty good metric for finding quality works without requiring a huge amount of views. (Despite what you'd think based on the Featured Box, Fimfiction actually has some pretty good taste!) The only thing it doesn't help with is new stories by people with a good number of followers, which always have weird ratios, and fandom classics, which have situations like Dangerous Business (currently a ratio of 1:2.7) and Fallout Equestia during its release which had more likes than views. I think that minimum threshold of 1000 views the right idea--even MLD "only" has 10K likes.
I think I'll take a look at Faun some time soon.
You do, huh? Then you might be interested in some statistics.
As of July 19, across all 87519 fics here are the Like/ Viewcount ratios:
Average Likes and views per story: 68:1034
Average Dislikes: 7
Or based on your system: 1:14
Averages skew towards the upper end of stories, so here is the median information.
Median Likes and views per story: 14:349
Median Dislikes: 3
Or based on your system: 1:21
I'm calculating the data for all stories including below 1000 views, so that may be skewing this data.
I'm currently in the midst of trying to update this blog post.
I thought it odd that you added all the votes, up and down, together, but when I checked this method out with my own stories, I found that counting just the up-votes didn't make a significant difference if I was rounding to whole numbers, or even a single decimal place, so your methodology seems reliable.
So... 1:6 is some sort of delimiter for excellence? Well, I'm happy I've got a story under that and one near the line, but is that based on a survey of agreed-upon "great" stories or something?
2329566
Reverse, views divided by votes.
2329716
YEAH BABY, WORK IT
2329801
Fascinating, thank you. :D
2329802
I don't think it's based on anything beyond "EQD pre-readers shooting the shit", really. :B 1:10 is certainly a useful standard, or even 1:14 or 1:21 as noted above.
I actually don't remember Fimfiction ever having a 5 star system. When was this changed?
2330443
Really, really old.
2329579 I think it might be the opposite, since later views would probably come from recommendations, rather than people browsing the site randomly. Follower count might also be a factor.
2329716 his presence is perfect.
2329801 how do you go about compiling these statistics?
2331578
To compile this data your going to need a few things, first, the raw data itself. The index.JSON that includes information that looks like this. That can be downloaded from the latest release of fimfarchive.
The tough part is programming in Java.
Here is the source code I wrote.
And here are some tips that will hopefully save you a few hours of googling to figure out Eclipse.
Just PM me if you run into any trouble.
Yay! My Harshwhinnial has a 6:1 views:votes ratio! I can write engaging fics!
Of course, if you count only upvotes, that skyrockets to 12:1.
I measured views:upvotes on my own fics at one point for similar reasons, and also to see whether there seemed to be a different engagement level based on various places the stories were featured. Results were inconclusive.